Mental diseases such as schizophrenia and depression are complex diseases for which susceptibility, development and treatment response are mediated by intricate genetic and environmental factors. Understanding the genetics of these diseases can illuminate significant insights into the development of diagnostics, pathogenesis and therapeutics of these diseases. With recent advancements in comprehensive genomic information and cost-effective genotyping technologies, genome-wide association studies (GWAS) have become a promising new tool for identifying modest genetic determinants of complex disorders. However, GWAS for psychiatric disorders have yet to bring definitive findings. Insufficient power to detect small-effect genes and inability to incorporate complex interactions are the two major attributes for lack of replicable findings. To ensure further success of GWAS, advanced analytical tools and strategies are needed to resolve these issues. This proposal intends to develop methodology in response to this need. Our long-term goal is to advance the efficacy of complex multimarker analysis and eventually to facilitate study design and marker selection. Modeling multimarker polymorphisms provides maximal amount of genomic information, and complex statistical modeling allows careful and collective consideration of potential genetic and environmental factors. In this proposal we focus on model-based haplotype analysis, and propose a two-stage framework to detect and to comprehend the association signals in GWAS. The first stage aims to effectively screen out regions with global haplotype-trait association. The second stage focuses on a more systematic examination of the specific patterns of haplotype effects. Motivated by issues arising in the collaborative works by the investigators, the central considerations of our methodology development include: (a) the efficient usage of haplotype information, (b) the formulation of regression-based framework, (c) the capacity to detect main and interaction effects, (d) a systematic inference progression from initial global screening to follow-up specific evaluation, and (e) the establishment of a solid theoretical foundation and robust implementation in user-friendly software. We will achieve our objectives through the following three specific aims: (1) to develop regression-based haplotype-similarity methods for detecting regions that exhibit genetic main and/or interaction effects, (2) to develop a penalized-likelihood regression approach for characterizing haplotypes of significant main and/or interaction effects within the identified regions, and (3) to apply the methods from aims (1) and (2) to the collaborative GWAS of mental disorders for method evaluation and disease gene detection, and to develop and distribute computer programs for public use. PUBLIC HEALTH RELEVANCE: Completion of the proposed work will provide effective statistical tools for a new process of studying the genetic etiology of complex diseases, from initial genome screening to subsequent explanatory examination. These tools can facilitate scientists'understanding of complex diseases and eventually lead to better design of prevention, detection and treatment strategies to improve human health.